Tissue-specific differences in cytosine methylation and their association with differential gene expression in sorghum

Abstract

It has been well established that DNA cytosine methylation plays essential regulatory roles in imprinting gene expression in endosperm, and hence normal embryonic development, in the model plant Arabidopsis (Arabidopsis thaliana). Nonetheless, the developmental role of this epigenetic marker in cereal crops remains largely unexplored. Here, we report for sorghum (Sorghum bicolor) differences in relative cytosine methylation levels and patterns at 5'-CCGG sites in seven tissues (endosperm, embryo, leaf, root, young inflorescence, anther, and ovary), and characterize a set of tissue-specific differentially methylated regions (TDMRs). We found that the most enriched TDMRs in sorghum are specific for the endosperm and are generated concomitantly but imbalanced by decrease versus increase in cytosine methylation at multiple 5'-CCGG sites across the genome. This leads to more extensive demethylation in the endosperm than in other tissues, where TDMRs are mainly tissue nonspecific rather than specific to a particular tissue. Accordingly, relative to endosperm, the other six tissues showed grossly similar levels though distinct patterns of cytosine methylation, presumably as a result of a similar extent of concomitant decrease versus increase in cytosine methylation that occurred at variable genomic loci. All four tested TDMRs were validated by bisulfite genomic sequencing. Diverse sequences were found to underlie the TDMRs, including those encoding various known-function or predicted proteins, transposable elements, and those bearing homology to putative imprinted genes in maize (Zea mays). We further found that the expression pattern of at least some genic TDMRs was correlated with its tissue-specific methylation state, implicating a developmental role of DNA methylation in regulating tissue-specific or -preferential gene expression in sorghum.

Figure 1.

Examples of MSAP profiles illustrating the various TDMRs (arrowed) across the seven studied sorghum tissues of a pair of reciprocal interstrain F1 hybrids (AB and BA) and its parental pure lines (A and B). Primer combinations used are EcoRI + ACG/HpaII (MspI) + TTA (A), EcoRI + AGG/HpaII (MspI) + TTG (B), EcoRI + AAC/HpaII (MspI) + TCT (C), and EcoRI + ACA/HpaII (MspI) + TTG (D).

Figure 2.

Methylation levels of the two major types, ^mCG and ^mCHG, calculated based on the MSAP profiles in five tissues of a pair of reciprocal sorghum interstrain F1 hybrids (AB and BA) and seven tissues in their parental pure lines (A and B). Asterisks (**) denote statistical significance at 0.01 level based on the Student’s t test.

Figure 3.

Hypo- and hypermethylation at CG or CHG of the randomly sampled 5′-CCGG sites by the MSAP marker in four (a pair of reciprocal F1 hybrids, AB and BA) or six (pure lines, A and B) sorghum tissues with reference to embryo.

Figure 4.

Pairwise comparison of methylation difference at CG or CHG of the randomly sampled 5′-CCGG sites by the MSAP marker among the five (a pair of reciprocal F1 hybrids, AB and BA) or seven (two pure lines, A and B) sorghum tissues.

Figure 5.

Relative distribution of the sorghum TDMRs based on the scored MSAP profiles of the five (a pair of reciprocal F1 hybrids, AB and BA) or seven (pure lines, A and B) studied tissues.

Figure 6.

Collective cytosine methylation values (in percentage) of CG, CHG, and CHH for each of the seven sorghum tissues. The tabulated values are based on the bisulfite sequencing data (given in Supplemental Fig. S2) for each of the four selected TDMRs: a low-copy retrotransposon (TDMR28), a putative sorghum protein kinase gene (TDMR40), a homolog of the maize endosperm NAM-related protein-coding gene (TDMR45), and a homolog of the maize fertilization-independent endosperm protein 1-coding gene (TDMR50).

Figure 7.

RT-PCR analysis of nine selected sorghum TDMRs bearing homology to TE (TDMR28) or protein-encoding genes. RT was performed on three batches of independently isolated RNAs from a pair of reciprocal F1 hybrids (AB and BA) and their parental pure lines (A and B). A sorghum actin gene (GenBank accession no. X79378) was used as a control for RNA input and DNA contamination (RNAs without RT). The gene product size and amplification cycles are labeled for each of the TDMRs.